SELECTION OF DATA 623 



laccoliths and sills that are well dissected, but not so far eroded that any 

 part is wholly lost. 



The Diagrams 



Figure 1 shows the location of average rocks of several families. The 

 nephelite syenites extend past the limits of the diagram far beyond the 

 location of the average. Siliceous pegmatities also extend at the opposite 

 corner beyond the limits shown. Most rocks, however, are well placed 

 in the figure. Eocks that are related seem in general to be near together 

 in the plot. The scale is different for the two coordinates, emphasizing 

 the effect of alkalies; but probably this is a proper emphasis and it gives 

 a better arrangement for study. 



The acid pegmatite selected was not the most siliceous rock known, 

 but was intended to show a more quartzose rock, than typical granite. 14 

 It was first thought that the average pegmatite would illustrate a siliceous 

 rock, but on averaging the 16 available analyses of granitic pegmatites 

 it was found that, as a group, pegmatite analyses show more alkalies and 

 less silica than granites. Probably those siliceous pegmatites so often 

 described are not analyzed, and the alkaline types therefore receive 

 emphasis quite unduly. 



Figure 2 shows a few of the larger series plotted in detail. For those 

 series which show a tendency to linear arrangement, lines are drawn on 

 the plots, but some of the series scatter so widely over the plot that a 

 linear connection is not justified. 



Figure 3 is based on a similar plotting, but with relatively small 

 amount of data for any district; in some cases only the analyses of the 

 two extremes. For all of these the individual points are here omitted 

 and only the linear tendency shown. A few which showed a tendency 

 to scatter at the end of a series are shown by a line split at the end. 



Figure 4 shows a number of supposedly related series of rocks in which 

 the geologic connection is less evident (gTadations were not found in 

 continuous exposure), but in which there appears a similar linear trend 

 in the series. In 10 or 15 other districts for which the writer has found 

 a quantitative study of the rocks, there have been suggestions of genetic 

 connection; but, since the various rocks are not connected in outcrop by 

 any gradational series and the diagram shows no pronounced linear trend, 

 they have been omitted from figure 4; they add nothing but the weight 



14 It Is found on page 77 of Washington's chemical analyses of igneous rocks, issued as 

 Professional Paper 99 of the United States Geological Survey. 1917. 



